Boot the kernel with "noapic" option. This will handle the clock timer
error message. As with the kapm-idled taking up all of the cpu cycles,
you'll just have to ignore it. There's nothing wrong with it, the new
kernel was coded this way intentionally. The new 2.4 kernels will just
assign all of the idled cpu time to kapm-idled. I forget the reason, but
there isn't any problem with it. If you look through the kernel
mailling list, you'll be able to read through the discussion of why
they did this. The clock timer error message, however is a problem. But
I don't know of any solutions yet. The APIC code doesn't seem to work
correctly with the VIA686a northbridge. You can recompile the kernel and
disable APIC support(It is enabled by default if you compile SMP, and
optional if you disable SMP). You are losing some performance by not
using APIC, but it's still better than not being able to use it at all.
On Mon, 23 Apr 2001, Tom Snell wrote:
> Yeah....next day when I cold-booted, the kapm-idled was gone, everything
> back to normal on the CPU. I'd been running the system for 16 hours
> when the kapm-idled began hogging the CPU cycles, so who knows? Why
> would an idle loop grab 80% of the CPU cycles if it's not even a
> process? I'm not savvy when it comes to kernel internals, so forgive me
> my stupification....;-)~
>
> The messages, though, continue.....I hadn't seen this in earlier 2.4.x
> kernel versions, but I've experienced very similar behavior after
> installing FreeBSD 4.1/4/2. Someone on a FBSD list told me it was a
> timing issue with AMD chipsets, correctable by recompiling the kernel
> without APM....of course, I couldn't stabilize the system enough to
> recompile! ;-) I hope Mandrake takes notice of this, since at least
> now we know we are not isolated cases. The error message itself seems
> to point to problems with VIA/AMD Athlon chipsets causing clock timing
> problems, so it's at least pointing to possible causes/solutions.
> Otherwise, if no fix is forthcoming, I either better install a good
> log-cleaning utility or jump back to 7.2! And that won't be easy,
> 'cause I finally got TuxRacer working! <G>
>
> Tom
>
> Will wrote:
> >
> > I see the messages too. It souds like the bug I was complaining about on the
> > list a few days ago (video dies when disk activity is high). Hopefully if
> > the new kernel is detecting the bug, it can work around it. We shall see.
> >
> > Kapm-idled is the idle loop in kernel 2.4.x, it's not a real process.
> >
> > On Monday 23 April 2001 00:55, you wrote:
> > > Install of 8.0 went well, and this version is really beautiful! But we
> > > do have some anomalies, perhaps with the new kernel 2.4.3, that did not
> > > show up in 2.4.1 or 2.4.2.
> > >
> > > From the time I first rebooted after install, stderr is sending the
> > > following lines to syslog (from dmesg):
> > >
> > > Apr 20 20:25:00 localhost kernel: Calibrating delay loop... <4>probable
> > > hardware bug:
> > > clock timer configuration lost - probably a VIA686a motherboard.
> > > Apr 20 20:25:00 localhost kernel: probable hardware bug: restoring chip
> > > configuration.
> > >
> > > This message repeats through the boot and into syslog during the
> > > session....syslog and /var/log/messages were up to 15MB in two days.
> > >
> > > At the same time, the CPU is getting creamed by the "kapm-idled" process
> > > (pid 3 on boot), anywhere from 51-85% of the cycles are being grabbed by
> > > this process. I don't know what this process is, but it looks like it
> > > has something to do with power management.
> > >
> > > At this point, if I confirm that the 2.4.3 kernel is having a problem
> > > with power management on motherboards with certain Athlon-class
> > > chipsets, I might recompile the kernel and take out APM, but I hope
> > > Mandrake can check this out and patch the kernel.
> > >
> > > FYI, I did not get these messages/CPU overload problem with kernels
> > > 2.2.18, 2.4.0 or 2.4.1.
> > >
> > > System Info:
> > >
> > > Athlon 650 MHz
> > > MSI-6195 mobo (AMD 756 chipset) [note: I had similar problems with
> > > FreeBSD 4.2 on this board)
> > > Corsair PC100 CAS2 SDRAM
> > >
> > > Appreciate if any Mandrake developers on the list check this out.
> > >
> > > Tom Snell
> >
> > --
> > Madness takes its toll; please have exact change
> >
> > Will Reinhart
> > [EMAIL PROTECTED]
>
John Kim
Linux System Engineer @ ASL - visit us at www.aslab.com